File and block storage have been used for decades and are well defined ways to store data. Object storage on the other hand is relatively new, hence the obvious question is – How different is Object storage? How is it beneficial? When can it be used? What benefits adds to the IT infrastructure? There are many related questions that need to be answered before implementing in the infrastructure. First the earlier storage system known as block and file storage should be understood, to understand the advantages and drawbacks of Object Storage.
What is meant by File Storage?
When storing, a file is structured and a file system is constructed. It is a way to arrange files so that the files can be stored and retrieved using the path. Other factors that need to be described are its owner; size and the file access are all stored as metadata in the defined file system. On a local network the best way to share files is Network Attached Storage popularly known as NAS. The file is shared securely among the users. It is really a nice way to share files over a local area network. However the file storage sharing on Wide Area network has its own problems. This is because file storage uses NAS box for sharing and managing files which includes locking and unlocking. Managing billions of file sharing over a wide area network like the world wide network (Internet) is a real challenge. The system is not built to handle such huge number of files. Such limitations only have come to light in recent years with the amount of shared files increasing exponentially. Of course NAS is really good system for sharing files in the surroundings it was build for, like local sharing where the number of files and the metadata is limited.
Meaning of Block Storage
When the limitations of file storage were known, block storage was invented for storing and sharing. Large business corporations have billions of files to share, which was not possible with file storage and NAS. In Block storage the data was divided into blocks of data. These blocks were later combined to form files as and when required. Every block has an address and the block can be reached by making a SCSI call. Block storage does not require metadata except for address. The block does not have the case of owner and is an array which can be accessed as per the applied needs. It is based on performance based on database stored at a granular level. Hence the management of billions of files is not a problem. However it also has the same limitations like file storage system when shared over a wider network. When the distance increases between users the performance is affected due to call latency. Hence block storage works very well over Storage Area Networks deployed by large business enterprises
Object Storage the new way of storage and sharing
Object storage combines the best of file storage and block storage. An object is a bundle which consists of data along with all its metadata. The object is then given an identification known as the object ID. The ID depends on what the object contains including files and metadata. The object can be retrieved by using the object ID.
Object storage uses flat structure as compared to block or file storage which is granular or layered. The object storage uses the idea of pool of objects. The storage of the objects can be anywhere geographically. Since the Object ID has flat address structure retrieval is the same for local files or files elsewhere over the globe. Limitations of using metadata are also not there. The assigning of protection level and type of application it has to be used can be decided without worry of performance going down. The object can be copied to multiple locations and if required can be stored locally. Thus object storage system rides over the problems with the file and Block storage systems.
Object storage is flexible in defining metadata as per the need and opens up the opportunities for wide area sharing. The richness of data is unaffected of number of files to be handled and also the distances between storage and use. It is the best system to manage service and storage of well defined data over a wide area network.
The development of object storage was done keeping in view the development of internet. The businesses knew they had to build storage infrastructure which was large and the cost had to be reduced. Thus object storage was based on cluster storage on servers with internal access. There are exceptions where object storage is used with RAID arrays, but the scaling of data is never a problem. It can be done simply by adding nodes. Protection of data is done by object replication on one or more nodes, with exceptions like use of erasure coding for data protection.
Hence Object storage solves the problems of increase in size and amount of data to be handled. The other storage systems face issues of durability and performance which object storage easily overcomes. It allows infinite scalability by simple addition of nodes and protection by replication and relocation. The cost of managing huge data is also reduced by object storage. So to sum up object storage is good for data handling over long distances with efficient. The only limitation is for transactional data file where it fails on consistency of data. The way to go is to use all these three storage systems at the levels they are made for – File Storage for local networks, Block Storage for files requiring security and Object storage for sharing files over the internet (Wide Area Network). Knowledge of all the storage is critical for the system administrator, as it helps to identify the storage system to be used. So the end note is to know and use as per the specific needs and usability features.